Search-based genetic optimization for deployment and reconfiguration of software in the cloud

Abstract

Migrating existing enterprise software to cloud platforms involves the comparison of competing cloud deployment options (CDOs). A CDO comprises a combination of a specific cloud environment, deployment architecture, and runtime reconfiguration rules for dynamic resource scaling. Our simulator CDOSim can evaluate CDOs, e.g., regarding response times and costs. However, the design space to be searched for wellsuited solutions is extremely huge. In this paper, we approach this optimization problem with the novel genetic algorithm CDOXplorer. It uses techniques of the search-based software engineering field and CDOSim to assess the fitness of CDOs. An experimental evaluation that employs, among others, the cloud environments Amazon EC2 and Microsoft Windows Azure, shows that CDOXplorer can find solutions that surpass those of other state-of-the-art techniques by up to 60%. Our experiment code and data and an implementation of CDOXplorer are available as open source software.

A Model-Driven DevOps Framework for QoS-Aware Cloud Applications

“…A tabu search (TS) heuristic has been used in [32] to derive component allocation under availability constraints in the context of embedded systems. Finally, Frey et al [33] proposed a combined metaheuristic-simulation approach based on a genetic algorithm to derive deployment architecture and runtime reconfiguration rules while moving a legacy application to the Cloud environment.…”

Cloud Modeling Languages by Example

“…It supports tbe cloud application and cloud environment perspective, though the focus with respect to cloud environments is mostly at the infrastructure and partly platform level for which constraints, pricing, and deployments can be specified. Dedicated tool support is offered by CloudMIG Xpress, which features automatic computation of optimal cloud-based deployments [26] and conformance checking of legacy software with respect to potential cloud providers [28].…”

“…In CAML, cloud offerings are annotated with costs mainly for the purpose of informing cloud consumers [19] and selecting cloud environments. CloudMIG covers pricing information of cloud environments for the optimization of software deployments [26]. RESERVOIR-ML deals with application performance indicators to acquire or release virtual machine instances depending on predefined elasticity rules.…”

Optimizing Multi-objective Evolutionary Algorithms to Enable Quality-Aware Software Provisioning

“…Multi-Objective Evolutionary Algorithms (MOEAs) [12], [36] is a class of search based approaches that addresses problems in which a decision maker aims at finding a solution that optimizes several conflicting objectives. MOEAs simulate population evolution to produce solutions exhibiting trade-offs between conflicting objectives such as grid jobs scheduler [16] as they are able to automate a set of configurations exploration [10]. A cloud infrastructure, characterized by its dynamic entities, is a typical example of self-adaptive system.…”

“…A cloud infrastructure can be abstracted by a set of software resources that run on top of Virtual Machines (VMs) dynamically starting/stopping in physical machines. MOEAs are thus used nowadays in several design case studies [16], [24], such as self-adaptive cloud scheduling problems, to maintain conflicting quality characteristics [12], [36] such as system performance, cost and safety. Beyond their applicability for cloud optimizers, MOEAs offer the following advantages to set-up autonomous self-adaptive engines working "at runtime": (i) no need for predefined solutions, (ii) the incremental optimization process can be stopped on-demand, (iii) operating multi-objective optimization and finding trade-offs.…”